工具条件未记录 / low / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / low / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
工具条件未记录 / medium / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / medium / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
工具条件未记录 / high / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / high / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
工具条件未记录 / xhigh / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / xhigh / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
工具条件未记录 / maximum / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / maximum / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
工具条件未记录 / max / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · 工具条件未记录 / max / 框架未记录 · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Benchmark definition and scores
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. has benchmark disclosures or scores across 5 model releases.
Benchmark versions and metrics
Aliases Factual error rate on difficult prompts (lower is better)
Responses with at least one factual error · % · Lower is better
This benchmark uses all publicly recorded configurations. Rankings are grouped by metric and evaluation protocol. Recorded tool access, budgets, and frameworks are shown separately. Scores remain available for reference when conditions are unspecified.
Responses with at least one factual error %
Tool access not recorded / low / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / low / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Tool access not recorded / medium / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / medium / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Tool access not recorded / high / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / high / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Tool access not recorded / xhigh / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / xhigh / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Tool access not recorded / maximum / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / maximum / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Tool access not recorded / max / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.
Factual error rate on difficult prompts · Internal error-inducing prompts in the GPT-6.1 Terra comparison article; exact dataset version, sample size and user-flagged provenance unspecified. · Responses with at least one factual error · Tool access not recorded / max / Framework not recorded · Factual error rate on difficult prompts (lower is better); y metric Answers with any factual error; x axis Cost (USD). Lower is better; article describes an internal benchmark of error-inducing prompts without identifying the user-flagged dataset.